13

A decade ago Alldredge et al. (1993) described a new class ofparticles, transparent exopolymeric particles (TEP). Transparentin seawater, they are visualized using a stain specific for poly-saccharides or mucoid matter, the Alcian blue (Gordon 1970;Wiebe and Pomeroy 1972). TEP range in size from 1 to 100 µmor more in diameter and have been found in high concentra-tion in different regions of the oceans (Alldredge et al. 1993;Passow and Alldredge 1994; Schuster and Herndl 1995; Mariand Kiørboe 1996; Niven et al. 1997; Kiørboe et al. 1998; Engel

2000; Mari et al. 2001; Beauvais et al. 2003) as well as in fresh-water (Worm and Søndergaard 1998; Berman and Viner-Mozzini 2001). They form by coagulation of dissolved or col-loidal organic matter (Mari and Burd 1998; Mari 1999; Passow2000). The discovery of TEP has challenged our understandingof biogeochemical processes in the pelagic zone. Owing totheir physicochemical properties, dynamics, and characteristicof distribution, TEP can affect trophic web structure and fluxprocesses in the ocean. There is both direct and indirect (i.e.,extrapolated from studies on colloids) evidence that TEP canact as sorption sites for dissolved organic (Schuster et al. 1998)and inorganic (Niven et al. 1997; Hirose and Tanoue 1998;Buffle et al. 1998; Schlekat et al. 1998; Schoemann et al. 1998;Schoemann et al. 2001) matter; attachment sites for bacteria(Alldredge et al. 1993; Mari and Kiørboe 1996); and a foodsource for microphageous protozoans (Shimeta 1993; Tranviket al. 1993), particle grazers such as copepods (Carman 1990;Decho and Moriarty 1990; Ling and Alldredge 2003), or passivefilter feeders such as tunicates (Flood et al. 1992). TEP canalso aggregate with other particles (Kiørboe and Hansen1993; Passow et al. 1994; Jackson 1995; Logan et al. 1995;

Production, concentration, and isolation of transparent exopolymericparticles using paramagnetic functionalized microspheresXavier Mari1* and Hans G. Dam2

1Marine Microbial Ecology Group, Observatoire Océanologique de Villefranche-sur-Mer, Laboratoire d’Océanographie deVillefranche-sur-Mer, Station Zoologique, B.P. 28, F-06234 Villefranche-sur-Mer cedex, France2Department of Marine Sciences, University of Connecticut, Groton, CT 06340-6048, USA

AbstractWe describe a new method to isolate and concentrate transparent exopolymeric particles (TEP), either natu-

rally occurring or produced by phytoplankton in cultures. The principles of this method are to use TEP-stickingproperties coupled with magnetic properties of paramagnetic microspheres to magnetically and selectively con-centrate TEP. First, we studied the formation of mixed aggregates of TEP and 1-µm microspheres using the bub-bling method. TEP precursors were obtained from diatom cultures and from natural seawater. Mixed aggregatesof TEP–microspheres were efficiently formed—all TEP formed were associated with microspheres, and no clus-ters of microspheres alone were observed. The number of attached microspheres scaled with TEP size raised toan exponent of 1.53. The second step was to combine TEP with paramagnetic microspheres and isolate and con-centrate the TEP by flowing out the solution containing mixed aggregates inside a strong magnetic field. About90% of the magnetically labeled TEP were trapped inside the magnetic field. TEP volume concentration meas-ured inside the magnetic field after isolation was about 5 times the initial TEP volume concentration. However,the predicted concentration factor, estimated from the relationship between number of attached microspheresversus TEP size, and the TEP size spectra in the initial solution was ∼9. This suggests that TEP observed after iso-lation inside the magnetic field may be compacted and/or that a fraction of the TEP pool inside the magneticfield may remain glued to the inner walls of the culture flask and, thus, the observed TEP size spectra could beunderestimated.

*Present address: IRD, Laboratoire d’Océanographie Dynamique et deClimatologie UMR 7617, Institut Pierre-Simon Laplace, Université Pierreet Marie Curie 4 place Jussieu, F-75252 Paris cedex 05, France. E-mail:mari@obs-vlfr.fr

AcknowledgmentsWe are grateful to K. Tang for assistance during the setup of the

experiments. Thanks are owed to Paul Wassmann for organizing thecruise in the Norwegian Fjords. This research was supported by a fellow-ship to X.M. from the commission of the European Communities (EVK3-CT99-00015) and a small-grant award from the Research Foundation ofthe University of Connecticut to H.G.D.

Limnol. Oceanogr.: Methods 2, 2004, 13–24© 2004, by the American Society of Limnology and Oceanography, Inc.

LIMNOLOGYand

OCEANOGRAPHY: METHODS

Jackson and Burd 1998; Mari and Burd 1998) into marine snowflocs and thus enhance the vertical flux of major elements, suchas carbon, and of adsorbed trace elements (Fowler and Knauer1986; Alldredge and Silver 1988; Burd et al. 2000). Finally, TEPcan modify the physical properties of seawater, such as vis-coelasticity (Jenkinson et al. 1998; Wyatt et al. 1998).

Two methods have been used so far to measure TEP, bothprimarily designed to observe TEP and describe their abun-dance and distribution using Alcian blue. First, the micro-scopic method (Alldredge et al. 1993; Passow and Alldredge1994) allows one to observe TEP directly under a compoundlight microscope, either by transferring the stained andretained TEP onto 0.2-µm polycarbonate filters onto micro-scope slides using the Filter-Transfer-Freeze technique (Hewesand Holm-Hansen 1983), or without transfer onto a slideusing brightfield microscopy (Logan et al. 1994). The micro-scopic method has mainly been applied to determine TEP sizespectra and to calculate TEP abundance and volume fraction.Second, the spectrophotometric method (Passow and All-dredge 1995) allows estimation of TEP abundance by quanti-fying Alcian blue staining. Both methods have advantages anddisadvantages (see review in Passow 2002), but the main dis-advantage for both is that neither one allows the study of TEPwithout staining and, thus, the possibility to work directlywith them.

A common feature of the two existing methods is theclose and intrinsic relationship between TEP and Alcian blue.Although no work has been published contesting the exis-tence of TEP, the use of this exclusive staining approach todefine a class of particles, which cannot been observed oth-erwise, could be criticized. Another property should be usednot only to detect and eventually quantify this pool oforganic matter, but also to isolate and concentrate them forfurther analysis. TEP have high sticking properties (Kiørboeand Hansen 1993; Dam and Drapeau 1995; Hansen andKiørboe 1997), i.e., high probability of adhesion upon colli-sion. The polysaccharides that compose TEP are highly surface-active (high stickiness) because of a high fraction of sulfatehalf-ester groups (Mopper et al. 1995) that promote the for-mation of divalent cations bridging. This high sticking prop-erty confers to TEP a major role for aggregation/sedimenta-tion processes, because the stickiness TEP is much higherthan that of cells (Kiørboe and Hansen 1993; Passow et al.1994; Logan et al. 1995), with values generally 2 to 4 ordersof magnitude higher than most marine particles. Because oftheir high stickiness, we reasoned that TEP could be detectedby tagging them with small inorganic particles, such asmicrometer-size microspheres. The first step for such anapproach would be to associate TEP and microspheres. Sincethe first description of the bubbling method by Baylor andSutcliffe (1963), this approach has been extensively used toproduce aggregates from dissolved organic matter (Johnsonet al. 1986; Kepkay and Johnson 1988; Kepkay 1991; Alberand Valieda 1994; Mopper et al. 1995; Mari 1999), as well as

producing mixed aggregates with bacteria, algae, and otherparticulate material (Blanchard 1989; Chen et al. 1998;Monahan and Dam 2001). During the present study, thebubbling method was used to produce mixed aggregates ofTEP–microspheres.

In the present study we describe a new method: (1) todetect TEP without Alcian blue staining and (2) to isolate andconcentrate TEP.

Materials and proceduresProduction of mixed aggregates of TEP–microspheres—Two sets

of experiments were conducted. First, we studied the abilityof TEP to form mixed aggregates with 1-µm PolybeadTM Poly-styrene Microspheres (Polysciences). To determine if theinteractions between micrometer size particles and TEPdepend on the chemical surface properties of particles, the fol-lowing 1-µm microspheres were used: (1) Polybead AminoMicrospheres, (2) Polybead Carboxylate Microspheres, and (3)Polybead Carboxy-Sulfate Microspheres. This study was firstconducted using TEP produced in the laboratory from exu-dates of diatom species (Thalassiosira weissflogii and Skele-tonema costatum) known to release large amounts of extracel-lular polymeric substances. Diatoms were grown in eightl-batch cultures on f/2 media with silica (Guillard 1975) atconstant illumination (ca. 150 µE m–2 s–1). Diatom cultures instationary growth phase (ca. 10-d-old culture) were filtered atlow and constant vacuum pressure (<100 mbar) through a47-mm-diameter GF/C Whatman filter (nominal pore size =1.2 µm). Second, because the physicochemical properties ofTEP produced from diatoms grown in cultures might differfrom naturally occurring TEP, the formation of mixed aggre-gates was also investigated during a spring cruise in three sub-arctic Norwegian fjords (Malangen fjord, Balsfjord, and Ulls-fjord). TEP were produced from natural seawater collectedusing a 5-L Niskin bottle at the maximum of fluorescencedepth on each sampling occasion. Seawater samples were directlyfiltered at low and constant vacuum pressure (<100 mbar)through a 47-mm-diameter GF/C Whatman filter, and the fil-trate was placed in a bubble adsorption column. This experi-ment was conducted once for each station.

For the first set of experiments (diatom cultures), a 4-Lbubble adsorption column made of borosilicate glass wasused to produce the mixed aggregates of TEP–microspheres.For the second set of experiments (natural seawater col-lected in Norwegian fjords), we used a 2.5-L bubbleadsorption column made of borosilicate glass. This differ-ence in column size was dictated by restriction of benchspace on board.

The microspheres were added to the bubbling column toobtain a final concentration (Cf) of 2.5 × 105 particles mL–1 (ca.25% of bacterial concentration). The volume of microspheressolution (Vi) to add was calculated as:

Vi = (CfVf)/Ci

Mari and Dam TEP production and purification

14

where Ci is the initial concentration of microspheres and Vf isthe volume of the bubbling column. The initial concentra-tion, Ci, was calculated from:

Ci = 6W1012/ρπd3

where W is the mass of polymer per mL of latex (0.025 g mL–1

for 2.5% solid volume according to manufacturer), ρ is thedensity of polymer in grams per mL (1.05 for polystyrene),and d is the microspheres’ diameter in micrometers. To avoidclumping prior to bubbling, the volume of microspheres toadd in the column (Vi) was diluted in 50 mL of 0.22 µm fil-tered seawater and sonicated for 5 min. The solutions con-taining the TEP precursors and the microspheres were mixedand then added to the column. The solutions were bubbledwith air at a constant gas flow rate of 100 mL min–1 (Kepkay1991). Bubbles were produced by a 10- to 20-µm pore size glassfrit. The air for bubbling passed through a 0.1-µm air filter(Polycap TF, Whatman) and a trap containing molecular sieves(4 Å, 45/60 mesh) for removal of impurities and moisture. The

solutions were bubbled for 90 min at room temperature (ca.20°C), and samples were collected in the middle of the col-umn at 10, 20, 30, 45, 60 and 90 min and immediately filteredfor TEP determination and microspheres enumeration. Duringbubbling the pH remained constant. Between each experi-ment, the column and the glass frit were soaked for 2 h with10% HCl and then rinsed three times with Milli-Q water.

TEP and microspheres determination—The microscopicmethod was used for quantifying TEP within the bubbling col-umn. TEP were stained with Alcian Blue and slides preparedfollowing the method of Passow et al. (1994). TEP size spectrawere determined by counting and sizing TEP at three succes-sive magnifications (Mari and Burd 1998). At least 10 imageswere taken per slide and for each magnification, and the TEPsize spectra were compiled by combining the size distributionsobtained at each magnification. The images were analyzedusing an image analyzing system (ImagePro Plus, MediaCy-bernetics). The number of 1-µm microspheres attached to TEPwas determined under visible light at 400× magnification, andindividual TEP were sized (Fig. 1). As a result, for the 9 exper-

Mari and Dam TEP production and purification

15

Fig. 1. Examples of TEP produced by bubbling, dissolved, and colloidal organic matter in the presence of 1-µm microspheres. The black dots associat-ed with TEP are the microspheres. The pictures were taken under a compound light microscope at 200× magnification.

iments conducted (equivalent to 48 sampling times), a total of1885 TEP were sized, and their attached microspherescounted. A relationship between TEP size and number ofattached microspheres was determined for each sample to esti-mate the fraction of total microspheres that was attached toTEP. The number of attached microspheres (n) per TEP was fit-ted to the power-law relationship:

n = adb

where d is the equivalent spherical TEP-diameter (ESD; µm),and a and b are constants for a given sample determined fromlinear regression of log-log transformed numbers of associatedmicrospheres versus TEP diameter (Mari and Kiørboe 1996).The fraction of attached microspheres was calculated by com-bining (i) the above relationship, (ii) the TEP size distribution,and (iii) the total concentration of microspheres in seawater.

TEP isolation and concentration using paramagnetic microspheres—The possibility of isolating and concentrating TEP using the

sticking properties of TEP (which allows the formation ofmixed aggregates of TEP–microspheres) coupled with the mag-netic properties of 1-µm paramagnetic carboxylated micros-pheres (PCM) was investigated. The principle of this method-ological approach is, first, to form mixed aggregates of TEPand magnetic microspheres and, second, to collect thesemixed aggregates using a magnetic field. We used encapsu-lated super-paramagnetic carboxylate microspheres (Estapor

Microspheres) of 0.95 µm diameter with high ferrite content(magnetic pigment content = 48%) and with high carboxylgroup density at the surface (COOH = 114 µeq g–1). These para-magnetic microspheres are characterized by a core-shell struc-ture (Polystyrene), where the iron oxide material does notinterfere with the surface components. They are colloidallystable in the absence of a magnetic field and total separationcan be achieved when PCM are exposed to a magnetic field,and they are easily resuspended once the magnetic field isremoved. These microspheres are used in many biotechnolog-ical applications, such as cell separation and nucleic acid iso-lation. To our knowledge this is the first time that PCM wereused for oceanographic applications.

An experiment was conducted using TEP colloidal precur-sors originating from two diatom cultures (Thalassiosira weiss-flogii and Skeletonema costatum). Diatoms were grown in 8-Lbatch cultures on f/2 media with silica (Guillard 1975) at con-stant illumination (ca. 150 µE m–2 s–1). Diatom cultures in sta-tionary growth phase (ca. 10-d-old culture) were filtered at lowand constant vacuum pressure (< 100 mbar) onto a 47 mmdiameter GF/C Whatman filter (nominal pore size = 1.2 µm).To obtain a full range of TEP volume concentrations, filtratesfrom S. costatum and T. weissflogii cultures were diluted withartificial seawater. The filtrate from S. costatum was diluted2- and 4-fold and the filtrate from T. weissflogii was diluted2-fold. As a result, five magnetic isolation experiments wereconducted (S. costatum, S. costatum 1:2, S. costatum 1:4, T. weiss-flogii, and T. weissflogii 1:2). TEP–PCM aggregates were createdby bubbling in a 2.5-L adsorption column (as above). PCMwere added in the bubbling column to obtain a final concen-tration (Cf) of 5 × 105 PCM mL–1, to reach optimum micro-spheres “demand” (100% attachment) and according to pre-diction (see Assessment, Fig. 5). During bubbling, the pHremained constant.

TEP were isolated and concentrated using the magneticproperties of the PCM. After 1 h of bubbling, mixed aggregatesof TEP–PCM were gradually concentrated by placing the sea-water from the bubbling column inside a magnetic field as fol-lows. A tissue culture flask was disposed in between two FlaskMagnets (Qiagen) held together by a rubber band (Fig. 2).Flask Magnets are 12.5 cm × 6 cm rectangular magnetic sepa-ration units specially designed for use with 250-mL tissue cul-ture flasks. Each unit consists of permanent magnets encasedin a plastic frame. Seawater from the bubbling column passedinside the tissue culture flask by means of a peristaltic pumpwhose optimal flow rate (25 mL min–1) depended on the flask

Mari and Dam TEP production and purification

16

Fig. 2. Experimental setup for isolation and concentration of TEP pro-duced by surface coagulation in the presence of PCM of 1 µm diameter

volume and the time necessary for complete separation of thePCM from the water. The seawater inlet was placed at the bot-tom of the flask while the seawater outlet was placed at thetop. Samples for microscopic TEP determination were col-lected in the initial solution (i.e., before magnetic separation),inside the tissue culture flask (i.e., after complete magneticseparation and resuspension of the trapped mixed aggregates),and in the final solution (i.e., solution supposedly “free” ofTEP), thereafter respectively called compartments A, B, and C.Because processing the solution takes about 100 min, the sys-tem was kept in a cooler filled with ice to minimize bacterialdegradation of TEP, once the mixed aggregates of TEP–PCMwere trapped within the magnetic field.

An experiment was conducted to verify the magnetic prop-erties of the PCM by running the same procedure, but dilut-ing PCM into Milli-Q water (final concentration = 2.5 × 105

PCM mL–1). A 2.5-L solution was processed through themagnetic field and the number of PCM in each compartment(A, B, and C) was enumerated. As a result, about 90% of thePCM were magnetically retained in compartment B, whichimplies that about 10% of the PCM were not magnetizable.

AssessmentRelationship between TEP size and number of attached

microspheres—No difference (ANCOVA, P > 0.05) was foundbetween treatments and between time points either in TEP for-mation or in abundance of microspheres per TEP size. That is,the relationship between the number of attached 1-µm micro-spheres and TEP size was independent of the TEP source (eitherThalassiosira weissflogii and Skeletonema costatum cultures ornatural seawater), bubbling time, or chemical properties of themicrospheres (Amino, Carboxylate, or Carboxy-Sulfate). The

number of 1-µm microspheres per TEP (n) scaled with TEPdiameter (d; µm) was

n = 0.91 d1.53

which indicates that the number of microspheres attached toTEP increases out of proportion to size (Fig. 3). As an example,10 microspheres of 1-µm should be attached to a TEP with anESD of 4.8 µm, whereas 100 microspheres should be attachedto a TEP with an ESD of 21.6 µm. However, because volumescales to L3, the volume-specific (n L–3) number of micro-spheres (n′) decreases with TEP size; that is, n′ = aL–1.57, wherethe value of the coefficient a depends on the shape used forcalculation volume. The independence of the scaling relation-ship from TEP source and microsphere surface property groupsuggests that the formation of mixed aggregates of TEP and1-µm microspheres is controlled by TEP sticking propertiesand size of the microspheres rather than the chemical charac-teristics of the microspheres. Additionally, mixed aggregates ofTEP–microspheres were always formed very efficiently, and100% of the TEP formed were associated with microspheres;no clusters of solely microspheres were observed.

Finally, the exponent of the relationship between the num-ber of particles attached to TEP and TEP size can be consideredas estimates of TEP fractal dimension (Mari and Kiørboe 1996),under the assumption that the number of attached particles isdirectly proportional to TEP porosity. An exponent of 1.53 forthe number of microspheres per TEP is consistent with fractaldimensions estimated for marine snow particles and TEPaggregates (1.39 to 1.52) and is consistent with dimension ofaggregates formed through shear coagulation (Logan andWilkinson 1990; Logan and Kilps 1995). Therefore, thismethod of aggregates formation with microspheres of knownsize could be used to estimate fractal dimension of aggregates.

Optimal initial microspheres concentration—For all experi-ments, ≥95% of the microspheres were attached to TEP after10 min bubbling. In addition, the percentage of attachedmicrospheres scaled with TEP volume concentration (Fig. 4)according to

%Microspheresattached = 9.8(TEP volume) + 95.5 .

This relationship implies that the theoretical percentageof attached microspheres increases when the TEP volumefraction increases. As the optimal percentage of attachedmicrospheres equals 100%, higher values indicate that TEPpresent in the solution are undersaturated in microspheresor that a fraction of the TEP pool is not associated withmicrospheres. Because in all experiments, no TEP were observedwithout attached microspheres, it is possible that some TEPwere undersaturated with respect to their microspheres’specific density. To reach the optimal %Microspheresattached

= 100% according to TEP volume concentration, the initialconcentration of microspheres has to be adjusted to the TEP

Mari and Dam TEP production and purification

17

Fig. 3. Number of 1-µm microspheres attached to a TEP (n) as a functionof its size (d, µm). Regression line was fitted to the observations (n = adb)

volume. The microspheres demand to reach 100% attachmentas a function of the TEP volume concentration was obtainedfrom the above relationship, considering that an increase byfactor x of the microspheres concentration amounts to divid-ing the %Microspheresattached by the same factor. Over a TEPvolume concentration range of 1 to 100 ppm, the optimal ini-tial concentration of microspheres to target maximumattachment varied between ca. 2.5 × 105 and 27.0 × 105

microspheres mL–1 (Fig. 5). Although volume concentrationof naturally occurring TEP can reach values higher than100 ppm (Passow et al. 1994; Mari and Burd 1998; Bermanand Viner-Mozzini 2001), they usually occured at concen-tration ≤20 ppm (Passow and Alldredge 1994; Mari andKiørboe 1996; Mari and Burd 1998; Kiørboe et al. 1998).Therefore, for an average TEP volume concentration of 20

ppm, an initial microspheres concentration of 5 × 105 to 10× 105 PCM mL–1 (optimum 7.5 × 105 PCM mL–1) is recom-mended (95% prediction interval).

TEP–PCM magnetizability—The probability of a mixedaggregate of TEP–PCM not being retained inside the magneticfield was estimated using the number of attached 1-µmmicrospheres versus TEP size relationship described above,and assuming that such an aggregate is not retained if >50%of the attached PCM are not magnetizable. Because onlyTEP–PCM aggregates between 1 and 3 µm diameter are poten-tially not retained due to the 10% nonmagnetizable PCM (Fig. 6),it is likely that the fraction of nonmagnetizable PCM does notsignificantly affect the overall magnetizability of theTEP–PCM pool. However, because a large fraction of the TEPpool is represented in the small size range, one would expectthis isolation procedure to be inappropriate to remove all TEPfrom solution, as a full TEP-size spectrum should be rapidlyreformed from remaining TEP precursors. Consequently,unless we are able to force aggregation of all TEP colloidal pre-cursors toward large-size classes prior to flowing out the solu-tion inside the magnetic field, compartment C should show asignificant TEP pool, because TEP free of PCM will be reformedbetween compartments B and C.

TEP–PCM isolation and concentration—During the five mag-netic isolation experiments, mixed aggregates of TEP–PCMpresent in the initial compartment (A) were efficientlyretained inside the magnetic field (Fig. 7). However, a rela-tively high TEP volume concentration was also observed incompartment C, with an average of 13.6 ± 10.3% of the initialTEP volume concentration (Table 1). Between compartmentsA and B, observed TEP volume concentration increased by anaverage factor of 5.02 ± 1.14. Expected TEP volume concen-trations in compartments B and C were estimated from (i) theinitial TEP size spectra in compartment A, (ii) the expectedfraction of attached PCM per size class considering the initial

Mari and Dam TEP production and purification

18

Fig. 4. Relationship between the percentages of attached 1-µm micro-spheres and TEP volume concentration in the bubbling column

Fig. 5. Predicted optimal initial concentration of 1-µm microspheres asa function of initial TEP volume concentration

Fig. 6. Probability of a TEP–PCM aggregate not being magnetizableaccording to its size

concentration of PCM and the PCM number versus TEP sizerelationship obtained above, and (iii) the fraction of nonmag-netizable PCM. With the assumption that mixed aggregates ofTEP–PCM should be retained if more than 50% of the attachedPCM are magnetizable, the expected TEP volume concentra-tion retained in B should be higher than what is observed withan average concentration factor of 8.77 ± 0.03, whereas itshould be lower in compartment C with values close to 0 ppm(Table 1). The discrepancy between observed and expectedTEP volume concentrations in compartment B might be anartifact caused by the compression of the particles due to highphysical forces generated by the magnetic field. In otherwords, TEP–PCM aggregates observed in compartment B couldbe compacted and, thus, more dense than those observed in A.

Such a process would lead to an underestimation of TEP vol-ume concentration in B. Although compacting of TEP intomore dense particles should not alter their chemical composi-tion, it may modify their staining capacity by reducing theirporosity. Therefore, such a modification of structure may alterthe determination of TEP concentration using the spec-trophotometric method (Passow and Alldredge 1995). Thisdiscrepancy may also be explained by the possibility thatsome TEP–PCM aggregates remained glued to the inner wallsof the culture flask. Considering that TEP–PCM aggregates set-tle in successive coats and stick to the walls thereby creating abiofilm-like structure, one might expect that a significantTEP–PCM fraction will not resuspend and will be lost frommicroscopic counts.

The study of the spectral slopes, δ, of observed andexpected TEP size spectra in the three compartments (Table 1)shows that δobserved and δexpected in compartment B do not differstatistically, whereas the spectral slopes in B are higher thanthe spectral slopes in C (either observed or expected) (P < 5%).This suggests that the fraction of small TEP is higher in com-partment C than in compartment B (i.e., large TEP–PCMaggregates are retained more efficiently inside the magneticfield that smaller ones). Additionally, δobserved and δexpected incompartment C were statistically different (P < 5%)—δexpected

being larger than δobserved—which suggests that the largeTEP–PCM aggregates observed were probably formed betweencompartments B and C. These two hypotheses are supportedby the discrepancies between observed and expected TEP vol-ume concentrations in compartments B and C.

Furthermore, the microsphere density inside TEP–PCMaggregates was determined in compartment C (Fig. 8). PCMdensity inside mixed aggregates was much lower than for TEP–microspheres aggregates initially formed (Fig. 3), and no clear

Mari and Dam TEP production and purification

19

Table 1. Observed and expected TEP volume concentrations and slopes, δ, of the TEP size spectra regression lines during the five magnetic purification experiments in different compartmentsa

T. weissflogii S. costatum S. costatum Average δT. weissflogii (1:2) S. costatum (1:2) (1:4) (±SD)

Observed

A δ –2.29 –2.41 –1.88 –2.21 –2.46 –2.25 ± 0.23

TEP volume (ppm) 89.7 34.8 95 38.4 12.3

B δ –2.49 –2.58 –2.78 –2.55 –2.27 –2.53 ± 0.18

TEP volume (ppm) 431.3 218 463.4 223.4 40.7

C δ –3 –3.31 –3.34 –4.1 –3.53 –3.46 ± 0.41

TEP volume (ppm) 12.4 10.9 7 3.1 0.9

Expected

B δ –2.26 –2.38 –1.84 –2.17 –2.41 –2.21 ± 0.23

TEP volume (ppm) 788.3 305.6 834.4 336.4 107.3

C δ –6.26 –6.41 –6.01 –6.22 –6.11 –6.20 ± 0.15

TEP volume (ppm) 0.01 0.01 0.01 0.01 0.01aExpected TEP size spectra in compartments B and C were estimated from (i) the initial TEP size spectra in compartment A, (ii) the expected fraction ofattached PCM per size class considering the initial concentration of PCM and the PCM number versus TEP size relationship (n = 0.91 d1.53), and (iii) thefraction of non-magnetizable PCM (i.e., 10%).

Fig. 7. Variations of the TEP volume concentrations during the magneticisolation experiments conducted with Skeletonema costatum andThalassiosira weissflogii exudates, in each compartment (A, B, and C)

relationship between TEP size and number of attached PCM wasobserved. This suggests that the TEP–PCM aggregates occurringin compartment C were probably formed between compart-ment B and C and that the few PCM observed in compartmentC and attached to TEP were the non-magnetizable ones.

Efficiency of the magnetic isolation procedure—The efficiencyof the magnetic isolation procedure to concentrate TEP wasestimated by calculating the fraction of the TEP pool mag-netically retained as a function of the initial TEP volume in1 mL (Fig. 9). For the five experiments conducted, the TEPvolume trapped inside the magnetic field (calculated as thedifference between TEP volumes in A and in C) averaged ca.88% of the initial TEP volume. However, the observed TEPvolume in B averaged 57.2 ± 12.6%, which suggests that30.9 ± 19.7% of the initial TEP volume observed in compart-ment A was missing. The same two processes detailed abovecould explain the missing TEP (i.e., a fraction lost via stickingon the walls of the culture flask and/or physical compressionof the TEP–PCM aggregates).

Volumes for compartments A and B—The experiments wereconducted with a volume of 2500 mL and of 250 mL, respec-tively for compartments A and B. Considering the efficiency ofthe magnetic procedure to concentrate TEP, this approachcould be used to obtain a “pure” solution of TEP (i.e., 100% invol/vol). A concentrated solution where TEP occupy the entirevolume of compartment B solution is obtained when the ratioof expected TEP volume magnetically retained:volume of Bequals 1. This can be achieved by either increasing the volumeof compartment A or decreasing the volume of compartmentB. A ratio of 1 was targeted by increasing the volume of com-

partment A from 2500 to 100 × 106 mL or decreasing the vol-ume of compartment B from 250 to 0.5 × 103 mL. Because theamount of TEP–PCM retained inside the magnetic fielddepends on the initial TEP concentration, the volume of thedifferent compartments was predicted over a range of TEP vol-ume concentrations from 5 to 200 ppm. The study of the sur-face plots of Log(Volume A) versus Log(TEP volume fraction)and of Log(Volume B) versus Log(TEP volume fraction) allowsdetermining the volumes for A or B for which the ratio,expected TEP volume magnetically retained:volume of B equals1 (i.e., Log[Ratio] = 0). For a fixed volume of 250 mL in com-partment B, it would be necessary to flow out ca. 1.8 × 106 mLand 56.2 × 106 mL to reach 250 mL of pure TEP in compart-ment B for initial TEP volume concentrations of 200 and 5 ppm,respectively (Fig. 10a). For a fixed volume of 2500 mL in com-partment A, it would be necessary to decrease the volume ofcompartment B to ca. 0.4 mL and 0.01 mL to obtain a pure TEPsolution in compartment B for initial TEP volume concentra-tions of 200 and 5 ppm, respectively (Fig. 10b). Because com-plete separation of TEP–PCM aggregates from seawater isachieved for an optimal flow rate of 25 mL min–1, it would take≥50 days to obtain 250 mL of “pure” TEP. Therefore, such a

Mari and Dam TEP production and purification

20

Fig. 8. Number of PCM attached to a TEP (n) as a function of its size(d, µm) in compartment C. The dotted line represents the regression line ofFig. 3

Fig. 9. Fraction of the TEP pool magnetically retained (in compartment B),not retained (in compartment C), and missing as a function of the initialTEP volume (mL)

“pure” TEP solution can only be obtained following this pro-cedure by decreasing the volume of compartment B.

DiscussionPhysical labeling of TEP—The first goal of this study was to

design a procedure to detect TEP without intermediary Alcianblue staining. We demonstrated that the presence of 1-µmmicrospheres clusters was due to the existence of an unstainedTEP matrix. Because all TEP formed in the presence of 1-µmmicrospheres were associated with microspheres and the 1-µmmicrospheres never aggregated by themselves by means of abiological glue insensitive to Alcian blue staining (no clustersof microspheres alone with a ghost matrix were observed), onecould use the association between micrometer-size spheresand TEP to physically label TEP. Although a large variety ofsurface-active substances (i.e., not only polysaccharides) maybe enclosed in the TEP formed, the systematic Alcian bluestaining of the microspheres clusters matrix suggests that thestained matrix could be operationally defined as TEP. Thisfinding has implications for TEP detection, because a cluster ofmicrospheres observed following a similar methodologicalapproach, but without using Alcian blue, can be defined as aTEP particle. Such an approach would free us from usingAlcian blue staining as an exclusive means to detect TEP. As anexample, a cluster of 100 1-µm microspheres should indicatethe presence of a TEP particle of ca. 22 µm equivalent spheri-cal diameter. This approach seems inappropriate to determinethe size of the TEP pool because it will be more time-consum-ing to enumerate the number of 1-µm microspheres in eachcluster than to determine the size via Alcian blue staining.

However, several other applications are possible. One examplewould be examination of TEP consumption by metazoans fol-lowing the ingestion of 1-µm fluorescently labeled micro-spheres (the size distribution of microspheres being altered bythe presence of TEP). Another is examination of TEP concen-tration on modification of trophic web structure.

Preparation prior to chemical analysis—The second goal ofthis study was to design a procedure to isolate and concentrateTEP. A current major limitation when working with TEP is theimpossibility to separate them from other particulate material,such as phytoplankters, bacteria, fecal pellets, and other parti-cles. We believe that the isolation procedure described herewill prove useful in several respects: e.g., in studies of the char-acterization and chemical composition of TEP or in studies ofthe ability of TEP to adsorb organic solutes, such as aminoacids (Schuster et al. 1998), and to complex with trace metals(e.g., Fe, Cd, and Mg; Schelkat et al. 1998; Schoemann et al.2001) and radioisotopes (e.g., 234Th; Niven et al. 1997).

Removing TEP and adsorbed substances—Due to their physico-chemical characteristics, TEP are suspected to adsorb efficientlymany dissolved compounds (Logan and Hunt 1987) such pro-teins, amino acids, fatty acids, and trace metals (Morel andGschwend 1987; Decho 1990; Wells and Goldberg 1991; Nivenet al. 1997; Schuster et al. 1998). Direct evidence for the adsorp-tion of a specific dissolved substance onto TEP can be obtainedby monitoring changes in TEP concentration as well as in thetargeted solute concentration in the water column. Thisapproach has been followed by Niven et al. (1997) in theirstudy of TEP as a strong organic ligand for Thorium-234. Theyshowed that 234Th removal was closely linked to variations ofthe TEP pool and concluded that TEP were efficient scavengersfor this radioisotope. We argue that the magnetic isolation pro-cedure can be used to directly investigate TEP adsorptive prop-erties. As an example, one could incubate a known concentra-tion of the targeted dissolved substance (e.g., 234Th) with TEP incompartment A and measure its concentration in compart-ment C after magnetic isolation; the missing quantity of solutebeing the amount lost via adsorption onto the TEP fractionthat has been trapped inside the magnetic field.

Comments and recommendationsPotential improvements—One of the limitations of the mag-

netic isolation procedure is that it does not allow the removalof all TEP and TEP colloidal precursors from solution. Just aswith icebergs, for which the emerged part represents only aminor fraction of the overall mass, a minor fraction of theoverall surface-active polysaccharide material exists as TEP(Chin et al. 1998). And again as with icebergs, removing theemerged part of the surface-active polysaccharide material pool(i.e., TEP pool) will be rapidly followed by a replenishment ofthe removed part. A way to improve the transfer efficiencybetween TEP colloidal precursors and the particulate phase ofthe TEP pool would be to introduce an additional nucleus foraggregation. Glass fibers have been used to promote the for-

Mari and Dam TEP production and purification

21

Fig. 10. Ratio of expected TEP volume magnetically retained to volume ofB as a function of TEP volume concentration and (a) volume of compart-ment A for a fixed volume of B = 250 mL, and (b) volume of compartmentB for a fixed volume of sample processed. (Volume of A) = 2500 mL.

mation of marine snow from Phaeocystis mucilage (Passow andWassmann 1994), and we hypothesize that adding such fibersduring the formation of TEP–PCM aggregates would signifi-cantly enhance aggregation processes. Another improvementcould come from the technique used to increase the frequencyof collision necessary for coagulation. The bubbling methodmight not be the most appropriate or efficient way of produc-ing TEP–PCM mixed aggregates (see Kepkay 1994 and Mona-han and Dam 2001). Other means for abiotic TEP formationfrom precursors, such as laminar (Passow 2000) and turbu-lent shear (Engel and Passow 2001) could be used and may givebetter transfer rates as the shear rate can easily be controlled.Finally, although the role of bacteria during the production,concentration, and isolation procedure was not investigated, itis likely that bacterial activity could alter the formation ofTEP–microspheres aggregates by degrading and solubilizing theaggregates (Smith et al. 1992), thereby diminishing the effi-ciency of this method for removing TEP from solution. Futureresearch should consider combining adsorption onto particles(such as glass fibers) and shear coagulation to improve effi-ciency of the magnetic isolation to remove TEP and investigatethe influence of bacteria on the isolation procedure.

ReferencesAlber, M., and I. Valieda. 1994. Production of microbial

organic aggregates from macrophyte-derived dissolvedorganic material. Limnol. Oceanogr. 39:37-50.

Alldredge, A. L., and M. W. Silver. 1988. Characteristics,dynamics and significance of marine snow. Prog. Oceanogr.20:41-82.

———, U. Passow, and B. E. Logan. 1993. The abundance andsignificance of a class of large, transparent organic particlesin the ocean. Deep-Sea Res. 40:1131-1140.

Baylor, E. R., and W. H. Sutcliffe. 1963. Dissolved organic mat-ter in seawater as a source for particulate food. Limnol.Oceanogr. 8:369-371.

Beauvais, S., M. L. Pedrotti, E. Villa, and R. Lemée. 2003. TEPdynamics in relation to trophic and hydrological condi-tions in the NW Mediterranean Sea. Mar. Ecol. Prog. Ser.262:97-109.

Berman, T., and Y. Viner-Mozzini. 2001. Abundance and char-acteristics of polysaccharide and proteinaceous particles inlake Kinneret. Aquat. Microb. Ecol. 24:255-264.

Blanchard, D. C. 1989. The ejection of drops from the sea andtheir enrichment with bacteria and other materials: areview. Estuaries 12:127-137.

Buffle, J., K. J. Wilkinson, S. Stoll, M. Filella, and J. Zhang.1998. A generalized description of aquatic colloidal interac-tions: the three-colloidal component approach. Environ.Sci. Technol. 32:2887-2899.

Burd, A. B., S. B. Moran, and G. A. Jackson. 2000. A coupledadsorption-aggregation model of POC/234Th ratio of marineparticles. Deep-Sea Res. 47:103-120.

Chen, Y. M., J. C. Liu, and Y.-H. Ju. 1998. Flotation removal of

algae from water. Coll. Surf. B: Biointerfaces 12:49-55.Chin W., M. W. Orellana, and P. Verdugo. 1998. Spontaneous

assembly of marine dissolved organic matter into polymergels. Nature 391:568-570.

Carman, K. R. 1990. Mechanisms of uptake of radioactivelabels by meiobenthic copepods during grazing experi-ments. Mar. Ecol. Prog. Ser. 68:71-83

Dam, H. G., and D. T. Drapeau. 1995. Coagulation efficiency,organic-matter glues and the dynamics of particles during aphytoplankton bloom in a mesocosm study. Deep-Sea Res.II. 42:111-123.

Decho, A. W. 1990. Microbial exopolymer secretions in oceanenvironments: their role(s) in food webs and marineprocesses. Oceanogr. Mar. Biol. Annu. Rev. 28:73-153.

———, and D. J. W. Moriarty. 1990. Bacterial exopolymer uti-lization by a harpacticoid copepod: a methodology andresults. Limnol. Oceanogr. 25:1039-1049.

Engel, A. 2000. The role of transparent exopolymer particles(TEP) in the increase in apparent particles stickiness a duringthe decline of a diatom bloom. J. Plankton Res. 22:485-497.

———, and U. Passow. 2001. Carbon and nitrogen content oftransparent exopolymer particles (TEP) in relation to theirAlcian Blue adsorption. Mar. Ecol. Prog. Ser. 219:1-10.

Flood, P. R., D. Deibel, and C. C. Morris. 1992. Filtration of col-loidal melanin from sea water by plankton tunicates.Nature 355:630-632.

Fowler, S. W., and G. A. Knauer. 1986. Role of large particles inthe transport of elements and organic compounds throughthe oceanic water column. Prog. Oceanogr. 16:147-194.

Gordon, D. C. 1970. A microscopic study of inorganic particlesin the North Atlantic Ocean. Deep-Sea Res. 17:175-185.

Guillard, R. L. 1975. Culture of phytoplankton for feedingmarine invertebrates, p. 29-60. In W. L. Smith, M. H. Chan-ley [eds.], Culture of marine invertebrate animals. Plenum.

Hansen, J. L. S., and T. Kiørboe. 1997. Quantifying interspe-cific coagulation efficiency of phytoplankton. Mar. Ecol.Prog. Ser. 159:75-79.

Hewes, C. D., and O. Holm-Hansen. 1983. A method for recov-ering nanoplankton from filters for identification with themicroscope: the Filter-Transfer-Freeze (FTF) technique. Lim-nol. Oceanogr. 28:389-394.

Hirose, K., and E. Tanoue. 1998. The vertical distribution ofthe strong organic ligand in particulate organic matter inthe North Pacific. Mar. Chem. 59:235-252.

Jackson, G. A. 1995. TEP and coagulation during a mesocosmexperiment. Deep-Sea Res. II 42:215-222.

———, and A. B. Burd. 1998. Aggregation in the marine envi-ronment. Environ. Sci. Technol. 32:2805-2814.

Jenkinson, I. R., T. Wyatt, and A. Malej. 1998. How viscoelas-tic effects of colloidal biopolymers modify rheologicalproperties of seawater. In I. Emri, R. Cvelbar [eds.], Proceed-ings of the 5th European Rheology Conference, Portoroz,Slovenia, Sept. 6–11, 1998; Prog. Trends Rheol. 5:57-58.

Johnson, B. D., X. L. Zhou, and P. J. Wangersky. 1986. Surface

Mari and Dam TEP production and purification

22

coagulation in seawater. Neth. J. Sea Res. 20:201-210.Kepkay, P. E. 1991. Surface coagulation and microbial respira-

tion in response to local advection and sea state in theNorth Atlantic. Mar. Ecol. Prog. Ser. 69:143-147.

———. 1994. Particle aggregation and biological reactivity ofcolloids. Mar. Ecol. Prog. Ser. 109:293-304.

———, and B. D. Johnson. 1988. Microbial response toorganic particle generation by surface coagulation in sea-water. Mar. Ecol. Prog. Ser. 48:193-198.

Kiørboe, T., and J. L. S. Hansen. 1993. Phytoplankton aggre-gate formation: observations of patterns and mechanismsof cell sticking and the significance of exopolymeric mate-rial. J. Plankton Res. 15:993-1018.

———, P. Tiselius, B. Mitchell-Innes, J. L. S. Hansen, A. W.Wisser, and X. Mari. 1998. Intensive aggregate formationwith low vertical flux during an upwelling induced diatombloom. Limnol. Oceanogr. 43:104-116.

Ling, S. C., and A. L. Alldredge. 2003. Does the marine cope-pod Calanus pacificus consume transparent exopolymerparticles (TEP)? J. Plankton Res. 25:507-515.

Logan, B. E., and D. B. Wilkinson. 1990. Fractal geometry ofmarine snow and other biological aggregates. Limnol.Oceanogr. 35:130-136.

———, H.-P. Grossart, and M. Simon. 1994. Direct observationof phytoplankton, TEP and aggregates on polycarbonate fil-ters using brightfield microscopy. J. Plankton Res. 16:1811-1815.

———, and J. R. Hunt. 1987. Advantages for microbes ofgrowth in permeable aggregates in marine systems. Limnol.Oceanogr. 32:1034-1048.

———, and J. R. Kilps. 1995. Fractal dimensions of aggregatesformed in different fluid mechanical environments. Wat.Res. 29:443-453.

———, U. Passow, A. L. Alldredge, H.-P. Grossart, and M.Simon. 1995. Rapid formation and sedimentation of largeaggregates is predictable from coagulation rates (half-lives)of transparent exopolymer particles (TEP). Deep-Sea Res. II.42:203-214.

Mari, X. 1999. Carbon content and C:N ratio of transparentexopolymeric particles (TEP) produced by bubbling exu-dates of diatoms. Mar. Ecol. Prog. Ser. 183:59-71.

———, and A. Burd. 1998. Seasonal size spectra of transparentexopolymeric particles (TEP) in a coastal sea and compari-son with those predicted using coagulation theory. Mar.Ecol. Prog. Ser. 163:63-76.

———, S. Beauvais, R. Lemée, and M. L. Pedrotti. 2001. Non-Redfield C:N ratio of transparent exopolymeric particles inthe northwestern Mediterranean Sea. Limnol. Oceanogr.46:1831-1836.

———, and T. Kiørboe. 1996. Abundance, size distribution andbacterial colonization of transparent exopolymeric particles(TEP) during spring in the Kattegtat. J. Plankton Res.18:969-986.

Monahan, E. C., and H. G. Dam. 2001. Bubbles: an estimate of

their role in the global oceanic flux of carbon. J. Geophys.Res. 105:9377-9383.

Mopper K., J. Zhou, K. S. Ramana, U. Passow, H. G. Dam, andD. T. Drapeau. 1995. The role of surface-active carbohy-drates in the flocculation of a diatom bloom in a meso-cosm. Deep-Sea Res. 42:47-73.

Morel, F. M. M., and P. M. Gschwend. 1987. The role of col-loids in the partitioning of solutes in natural waters, p. 405-422. In W. Stumm [ed.], Aquatic surface chemistry. Wiley.

Niven, S. E. H., P. E. Kepkay, and J. B. C. Bugden. 1997. Therole of TEP in 234Th scavenging during a coastal diatombloom. Radioprotection-Colloques 32:213-218.

Passow, U. 2000. Formation of transparent exopolymer parti-cles, TEP, from dissolved precursor material. Mar. Ecol.Prog. Ser. 192:1-11.

———. 2002. Transparent exopolymer particles (TEP) inaquatic environments. Prog. Oceanogr. 55:287-333.

———, and A. L. Alldredge. 1994. Distribution, size and bac-terial colonization of transparent exopolymer particles(TEP) in the ocean. Mar. Ecol. Prog. Ser. 113:185-198.

———, A. L. Alldredge, and B. E. Logan. 1994. The role of par-ticulate carbohydrate exudates in the flocculation ofdiatom blooms. Deep-Sea Res. I 41:335-357.

———, and A. L. Alldredge. 1995. A dye-binding assay for thespectrophotometric measurement of transparent exopoly-meric particles (TEP). Limnol. Oceanogr. 40:1326-1335.

———, and P. Wassmann. 1994. On the trophic fate of Phaeo-cystis pouchetii (Hariot): The formation of marine snow by P.pouchetii. Mar. Ecol. Prog. Ser. 104:151-163.

Schlekat, C. E., A. W. Decho, and G. T. Chandler. 1998. Sorp-tion of cadmium to bacterial extracellular polymeric sedi-ment coatings under estuarine conditions. Environ. Toxi-col. Chem. 17:1867-1874.

Schuster, S., and G. J. Herndl. 1995. Formation and signifi-cance of transparent exopolymeric particles in the northernAdriatic Sea. Mar. Ecol. Prog. Ser. 124:227-236.

———, J. M. Arrieta, and G. J. Herndl. 1998. Adsorption of dis-solved free amino acids on colloidal DOM enhances col-loidal DOM utilization but reduces amino acid uptake byorders of magnitude in marine bacterioplankton. Mar. Ecol.Prog. Ser. 166:99-108.

Schoemann, V., H. J. W. de Baar, J. T. M. de Jong, C. Lancelot.1998. Effects of phytoplankton blooms on the cycling ofmanganese and iron in coastal waters. Limnol. Oceanogr.43:1427-1441.

———, R. Wollast, L. Chou, and C. Lancelot. 2001. Effects ofphotosynthesis on the accumulation of Mn and Fe byPhaeocystis colonies. Limnol. Oceanogr. 46:1065-1076.

Shimeta, J. 1993. Diffusional encounter of submicrometer par-ticles and small cells by suspension feeders. Limnol.Oceanogr. 38:456-465.

Smith, D. C., M. Simon, A. L. Alldredge, and F. Azam. 1992.Intense hydrolytic enzyme activity on marine aggregatesand implications for rapid particle dissolution. Nature

Mari and Dam TEP production and purification

23

359:139-141.Tranvik, C. J., E. B. Sherr, and B. F. Sherr. 1993. Uptake and uti-

lization of ‘colloidal DOM’ by heterotrophic flagellates inseawater. Mar. Ecol. Prog. Ser. 92:301-309.

Wells, M. L., and E. D. Goldberg. 1991. Occurrence of smallcolloids in sea water. Nature 353:342-344.

Wiebe, J. W., and L. R. Pomeroy. 1972. Microorganisms andtheir association with aggregates and detritus in the sea: amicroscopic study. Mem. Ist. Ital. Hydrobiol. 29:325-352.

Worm, J., and M. Søndergaard. 1998. Alcian Blue-stained par-

ticles in an eutrophic lake. J. Plankton Res. 20:179-186.Wyatt, T., I. R. Jenkinson, and A. Malej. 1998. How viscoelastic

properties of colloids, transparent exopolymeric particlesand marine organic aggregates, modify turbulence andplankton biodynamics in the sea. In I. Emri, R. Cvelbar [eds],Proceedings 5th European Rheology Conference, Portoroz,Slovenia, Sept. 6–11, 1998; Prog. Trends Rheol. 5:65-66.

Submitted 11 July 2003Revised 14 September 2003

Accepted 30 November 2003

Mari and Dam TEP production and purification

24